LaOT [Adapter]: benchmark results
Rank-16 LoRA adapter for HyperbeeAI/Tulpar-7b-v1, targeting gate_proj, down_proj and up_proj, with alpha 16 and dropout 0.05. Adapter weights remain separate from the base-model parameter total. Provider: Other. Access: Open.
Aggregate Rating 1324 ± 36, rank #8262 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - HellaSwag | 78.96 | Normalized accuracy (%) (10-shot) | 36.3 |
| Open LLM Leaderboard v1 - WinoGrande | 74.11 | Accuracy (%) (5-shot) | 35 |
| Open LLM Leaderboard v1 - ARC Challenge | 55.63 | Normalized accuracy (%) (25-shot) | 33.8 |
| Open LLM Leaderboard v1 - MMLU | 50.3 | Accuracy (%) (5-shot) | 29.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 44.72 | MC2 (%) (0-shot) | 28.2 |
| Open LLM Leaderboard v1 - GSM8K | 0 | Accuracy (%) (5-shot) | 4.5 |
Interactive version: theaggregate.ai/model?slug=laot-adapter · How It Works · Data refreshed daily, snapshot 2026-11-08.